Crane Load Spectrum Based On Support Vector Machine (svm) Method For Study

In recent years, with the rapid development of national economy andinvesting heavily in infrastructure, the major technical equipment industry isbooming, and crane occupies a large proportion in major technical equipmentindustry, Since the crane is heavy equipment mostly, in the event of accidents,economic losses heavily, which can easily result in casualties. Therefore, ourcountry had placed enormous demands for using cranes safely, and listed thecrane as special equipment. Through the survey in crane accident, found that itsprimary cause is the endurance failure. So, most government and the scientificresearch institution began to research the crane endurance failure questionmassively, Hope that report the user before equipment endurance failure.This paper resrarch the metal fatigue fracture related theory, know thatprecondition for solving the crane endurance failure is that establishing realservice condition of simulating the crane metal structure, the representativetypical load--time course, namely load spectrum. As a result of load randomnessand the uncertainty, the actual result is unable to apply directly in the theoreticalanalysis and the project practice, but need to construct equivalent load spectrumwhich can reflect naturally the time variation of the crane structure system undereach kind of operating mode.This paper take the casting overhead travailing crane and general overheadtravailing crane as the research object, through investigation, collect a part ofdata sample of crane working condition. and using the support vector machinelinear regression theory based on the statistical learning method to train thesample data, establishes the corresponding types of bridge crane the nonlinearmapping between relation operation cycles and lifting load, using the mappingrelationship, can predict the corresponding types or unknown bridge craneequivalent load spectrum. And making the software about obtaining andpredicting the crane equivalent load spectrum by use of visual programminglanguage VC++6.0. The software is applied to the engineering examples, andcomparing the calculation results with actual results, show a higher practicability and coincide. And comparing this method with the least squaremethod and neural network method, using the same sample data to obtain andpredict the crane load spectrum, proved the advantages of support vectormachine (SVM) method. This software operation is simple and clear, operatorscan use it not need to have the support vector machine related knowledge. Themost important is that the research resules lay the foundation for the follow-upsoftware development about crane fatigue life prediction.